Fusion of Human Demonstrations for Automatic Recovery during Industrial Assembly

Abstract

A novel approach for recovering from errors during automated assembly in typical mating operations is presented. It is based on automated error detection w.r.t. a predefined process model, followed by choosing a recovery strategy from an optimized repository. The latter comprises successful strategies that were recorded from human demonstration during a large scale user study. This paper shows how to enhance the process model with additional data, how to record new strategies in case where no suitable strategy is found, how to optimize a set of strategies, and how to select the most appropriate recovering strategy. A particular focus is the fusion of various human demonstrations in order to optimize them. The added value of the new approach is demonstrated by an experimental validation.